mlx-vlm Integrates LFM2.5 DSpark for 3.7x Speedup on M5 Max
helloiamleonie · x · 2026-08-22
mlx-vlm added support for Liquid AI's LFM2.5 DSpark in v0.6.16. Using exact speculative decoding, it achieves up to 3.7x faster generation on M5 Max with zero output drift. The implementation includes shared Metal GEMV kernels and a hybrid-capture rollback mechanism to ensure exact parity for both dense and MoE models.
More from Infra
- Muse Spark Open runs locally on a Mac M3 Ultra via llama.cpp: frontier model on desktop — pcuenq · 2026-10-16
- Muse-Spark open-weights drop: 816B-A42B MoE model gets 359GB MLX 3-bit build — ns123abc · 2026-10-16
- vLLM ships Day-0 LoRA RL for Muse Spark Open, lifting AIME from 0.56 to 0.85 — vllm_project · 2026-10-16
- DSpark draft model for Muse Spark Open proposes 7 tokens per pass with 23-42% acceptance — vllm_project · 2026-10-16
- Cloudflare Stream's WebRTC ultra-low-latency live streaming is now Generally Available — ritakozlov · 2026-10-16
- YC-backed Overshoot launches inference optimization, claims $5M saved for AI teams — ycombinator · 2026-10-16